Table 2.

The existing DL models on inferior alveolar nerve and their function and performance.

Authors	DL Models	Year	Training Dataset	Validation/Test Dataset	Functions	Best Performance of DL	Time- Consuming
Grana et al. [35]	CNN	2022	68	8 for validation 15 for test	IAN detection	IoU: 0.45 DSC: 0.62	No
Kaski et al. [36]	CNN	2020	128		IAN detection	Precision: 0.85 Recall: 0.64 DSC: 0.6 (roughly)	No
Song et al. [37]	CNN	2021	83	50	IAN detection	0.58 ± 0.08	86.4 ± 61.8 s
Hwang et al. [38]	3D U-Net	2020	102		IAN detection	Background accuracy: 0.999 Mandibular canal accuracy: 0.927 Global accuracy: 0.999 IoU: 0.577	No
Nalampang et al. [39]	CNN	2022	882	100 for validation 150 for test	IAN detection	Accuracy: 0.99	No
Jacobs et al. [40]	CNN	2022	166	30 for validation 39 for test	IAN detection, relationship between IAN and the third molar	Precision: 0.782 Recall: 0.792 Accuracy: 0.999 DSC: 0.774 IoU: 0.636 HD: 0.705	21.2 ± 2.79 s
Fu et al. [41]	CNN	2022	154	30 for validation 45 for test	IAN detection, relationship between IAN and the third molar	The third molar Accuracy: 0.9726 DSC: 0.9730 IoU: 0.9606 Mandibular canal Accuracy: 0.9563 DSC: 0.9248 IoU: 0.9003	6.1 ± 1.0 s for segmentation 7.4 ± 1.0 s for classifying relation
Yi et al. [42]	Canal-Net	2022	30	20 for validation 20 for test	IAN detection	Precision: 0.89 ± 0.06 Recall: 0.88 ± 0.06 DSC: 0.87 ± 0.05 Jaccard index: 0.80 ± 0.06 Mean curve distance: 0.62 ± 0.10 Volume of error: 0.10 ± 0.04 Relative volume difference: 0.14 ± 0.04	No
Shin et al. [43]	CNN	2022	400	500	IAN detection	Precision: 0.69 Recall: 0.832 DSC: 0.751 F1 score: 0.759 IoU: 0.795	No